Johnson et al. are another attempt to replace guesswork and unfalsifiable storytelling by quantitative and testable results. („Man kann zwar ohne Mathematik einen Gegenstand sehr geistreich besprechen, man kann aber nicht ohne Mathematik etwas Zusammenhängendes in innerlich zusammenhaltender Weise überdenken. Unerquicklich und unzusammenhängend sind die Abschnitte, in denen die Mathematik noch nicht Fuß gefaßt hat.“; Georg Simon Ohm) What I dislike about this model is, that it violates the most basic rule of statistical analysis. Whenever a result depends only on one or two points at the extremes of the data spread and changes substantially when eliminating those, it is not to be considered valid. Here the whole model hinges on the first and second data point entirely without guaranteeing, that the very first is even noticed.
Cohen et al.‘s modelling is rather unconvincing and the authors say so themselves. Looking at the real experiences in East Germany and their support for students with children there may well be something in it. There is a positive side to the result too. It implies that those with children not being educated is not the same as those with children being unable to be educated. There seems to be general agreement now, about the theory of Eugenics being totally discredited, but all these politically correct assertions sound much more like public pledges of faith to me, than like reports of testable scientific results. There is a huge observable difference between groups with and groups without children in our society and so far I don’t see the predictions from the end of the nineteenth century convincingly falsified anywhere.
This week’s story by Gordon Cash is somewhat surprising. It is so unconvincing as to not make sense even as a fantasy – it would be different if the Christian fundamentalists were replaced by Muslims. So life is different on the other side of the Atlantic.
The moment a once overbearing orthodoxy proscribing all dissent is finally discredited, loads of timid conformists come out of hiding to kick it when it’s down. Unfortunately it will be studies like Taylor et al. that get quoted in future years and not their derided and ignored precursors from years past.
Here’s the link to this week’s complete list.